It's DONE: Direct ONE-shot learning with quantile weight imprinting

TL;DR

This paper introduces DONE, a simple, nonparametric one-shot learning method that adds new classes to pretrained neural networks without retraining or modification, inspired by Hebbian theory and using quantile normalization.

Contribution

DONE is a novel one-shot learning approach that directly imprints neural activity as weights, avoiding retraining and interference with existing classes.

Findings

DONE achieves decent accuracy with current pretrained models.

The method is simple, deterministic, and hyperparameter-free.

It effectively adds new classes without retraining the entire network.

Abstract

Learning a new concept from one example is a superior function of the human brain and it is drawing attention in the field of machine learning as a one-shot learning task. In this paper, we propose one of the simplest methods for this task with a nonparametric weight imprinting, named Direct ONE-shot learning (DONE). DONE adds new classes to a pretrained deep neural network (DNN) classifier with neither training optimization nor pretrained-DNN modification. DONE is inspired by Hebbian theory and directly uses the neural activity input of the final dense layer obtained from data that belongs to the new additional class as the synaptic weight with a newly-provided-output neuron for the new class, transforming all statistical properties of the neural activity into those of synaptic weight by quantile normalization. DONE requires just one inference for learning a new concept and its procedure is simple, deterministic, not requiring parameter tuning and hyperparameters. DONE overcomes a severe problem of existing weight imprinting methods that DNN-dependently interfere with the classification of original-class images. The performance of DONE depends entirely on the pretrained DNN model used as a backbone model, and we confirmed that DONE with current well-trained backbone models perform at a decent accuracy.